Dear Statalisters,
I am working with a panel data set where I believe one of my x
variables to be endogenous and that there is a possible simultaneity
issue. I have been reading about panel GMM methods and I am hoping
that one of these techniques may be appropriate for my situation. My
strategy is to difference the data to control for any constant
unobserved characteristics and use lagged explanatory variables as
instruments. The majority of the readings that I have come across
address dynamic panels, which leaves me with a few questions:
Are there situations where the weak exogeneity assumption is not
valid? I believe that my regressor is contemporaneously correlated
with the error term, and from what I understand, I can use lagged t-2
explanatory variables as instruments (Ziliak 1997).
Are there any tests that need to be performed before I try to estimate
one of these models? For autocorrelation or the like?
How does one choose between a static versus a dynamic panel?
Any suggestions or pointers to appropriate references would be greatly
appreciated. Many thanks to you all!
Regards,
Susan
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
